In this Paper we investigate the performance of the five efficiency estimation methods which include the stochastic frontier model estimated by maximum likelihood (SFML), the stochastic frontier model estimated by corrected ordinary least squares (SFCOLS), the data envelopment analysis (DIA) model, the combined estimation of SFML and DEA (SFML + DEA), and the combined estimation of SFCOLS arid DIA (SFCOLS+ DEA) using Monte Carlo analysis. The results include: 1) SFML provides most accurate efficiency estimates for the sample sloe 150 or over,2) SFML+DEAor SFCOLS + DIA Perform better for the cases with sample sloe 25, 50, and low random errors, 3) SFCOLS performs better for the close with sample sloe 25, 50, and very high random errors.